This year I once again had the good fortune to be able to attend some of The Insurance Network's Technology conference in London, or TINTech as it's colloquially known. Once again, it was a well-organised event, with senior leaders from the insurance industry choosing between various tracks. This year rather than choosing between the line of business focused tracks I opted to look attend those focused on some of the hot and new technologies.
As a result my morning was occupied with the topic of AI with the content and discussions ranging across use of AI, use cases, data science, machine learning and data stewardship. The speakers were at the data scientist end of the AI adopter scale, demonstrating strong domain knowledge and a preference for tools and approaches that allowed control over the algorithms. The discussion through the morning yielded some interesting insights:
- Adoption of AI is advanced in insurance; however, deployment of AI is not widely distributed. Celent's own research suggests that many mid to large size insurers are actively leveraging AI but not evenly across the whole value chain. As discussed at TINtech, access to skills and tools is limiting this.
- A data scientist may not be the right guy to talk to a startup. This is a perhaps ungracious observation but one worth making all the same. There were attendees in the audience who were actively engaging with the start up community to leverage AI, while the data scientists in attendance seemed to prefer a self build approach and an understanding of the detail of the solution.
- While the experience at TINtech is anecdotal, it does follow a pattern of different types of adopters - those that prefer open source tools with greater control, those that leverage enterprise solutions for increased speed and the citizen data scientists with little training looking for tools to help them leverage AI. There are pros and cons to each approach. I predict some tension between these groups in the future as we will see insurers growing teams of all these types. Key takeaways: there isn't one right solution for all problems, nor one right approach. Trial and error will continue to play a role in insurers adoption of AI.
- AI is necessary to grow, but humans are still needed. Historically growing meant more people. For one panelist, double digits growth in one territory meant that having abundant capacity of machines doing some of the work was the only way to keep pace with growth. On the question of will robots take our jobs however, the panel agreed that no one gets worried when machines do work humans don't want to do.
In the afternoon, I took the opportunity to see the update on the activity at B3i. 2018 is already proving to be the year of blockchain in insurance, with the recent launch of Insurwave, among others. While blockchain has been referred to as a solution looking for a problem in insurance, I find myself coming around to the technology. There was concern expressed in the audience: Given the competing blockchain and DLT platforms was insurance facing a VHS vs Betamax choice? The response was, much like in AI, it is relatively inexpensive to trial the technology -- and to not do so would put you behind your competitors.
This blog entry has been reposted with permission
